AI Safeguarding Must Begin Before the Image Is Generated

A photograph taken at a school event, youth service, graduation or ministry outreach may appear harmless. In an age of generative artificial intelligence, the same image can be copied, altered, sexualised, impersonated and circulated without the subject's knowledge or consent.

The technological act may be synthetic. The humiliation, fear, reputational damage, safeguarding risk and loss of trust are not.

Educational and Christian institutions therefore need a larger understanding of AI governance. It is not enough to publish rules about student essays or ask staff to verify generated text. The institution must govern the images it collects, the systems people can access, the conduct it prohibits, the route through which concerns are reported, the way evidence is protected and the care offered to a person who has been harmed.

AI safeguarding must begin before the image is generated.

On 23 July, ABC News reported that police were investigating alleged AI-generated explicit material discovered in a former teacher's school account at an Australian school. According to the school's account, access to the material was restricted, the material was secured and handed to police, the staff member was dismissed after an internal investigation and relevant regulators were notified. The school said it had no evidence that the material had been shared.

These remain allegations within an investigation. A news report cannot establish guilt, determine the exact method of…